I know they're not using puzzles from this thread for TT puzzles, but I thought this was cool enough that I'd post it anyway. It's from a 960 game, but I think it's still workable... and since it won't be used, I feel no obligation to transpose the problem to a regular game.
http://www.chess.com/echess/game?id=59054386
homernh, white can save the bishop http://www.chess.com/tactics/server?id=223393
Thanks, I obviously didn't check that with an engine.
How about this puzzle?:
What do you think? Also checked with an engine
doesn't 3. Qg8 cxd4 4. Kxd4+ Ka3 make Qg8 a bad move??? Also, if the knight takes the Q, can't blacks pawn take whites Q and get promoted?
doesn't 3. Qg8 cxd4 4. Kxd4+ Ka3 make Qg8 a bad move??? Also, if the knight takes the Q, can't blacks pawn take whites Q and get promoted?
Please look at the coordinates on the board...
Your turn 
Doesn't the switchable move order make it bad for tactics trainer?
It makes it unworkable for TT.
Hmmm, a 15 move TT puzzle, not likely :P But seriously, that puzzle isn't very TTesque either.
Sure, the first four moves are forced, but the rest of them are not. Maybe the first four would be great for TT with analysis explaining why, although someone who sees the first 4 would probably already understand.
Out of curiosity, how much of this can be called a tactic and how much of it is positional in nature? You know, a "Positional Trainer" could be a good addition to chess.com. Porbably a much harder thing to make and do well.
thanks Scottrf , it's my first attempt